This is a screenshot of Oracle VM manager.
So we have anti-affinity group from the Perspective of dropbox menu. And then we can create anti-affinity group. And then we can create the group by providing name. And which virtual machines must be placed to physically different location? You assign virtual machines to anti-affinity group. That’s what you can do.
So finally you have a newly created anti-affinity group. In this example, only one VM. And you can assign additional VMs as well.
And also, server poor policy– so you can also manage VMs policy-based. So first policy is distributed resource scheduling. It dynamically manages quality of service based on policy. And also, it balances server load, because VMs running out of heavily loaded servers can be migrated to least-loaded server. By doing so, we can balance the load.
Behind the scenes, DRS uses live migration feature. As you can see, when you choose a pool server pool name from Perspective dropbox menu, you can select Policies. And in this example, policy type is DRS– Distributed Resource Scheduling.
We monitor activities for 10 minutes. At an average for 10-minute time period, server processor threshold, 90%. So your actual load is greater than 90%. Then it’s possible to move a VM from one VM server to another VM server. So you can edit it, and you can create new policy.
So the last topic is the server pool clusters.The cluster features are provided by capability of OCFS2, which is a cluster file system for Linux, and HA functionality in Oracle VM. The cluster option offers shared access to the pool resource– the resources residing in the OCFS2 file system– and also HA for the virtual machines running in the server pool.
In the management node, Oracle VM Manager is up and running, and the computer servers are clustered. Cn0 1, 2, 3, that’s a generic name for this example. And also, VM instances are running out of compute node. And here we have a default server pool, Rack1_ServerPool. And when you look at the storage layout, in the ZFS storage appliance, it had a preconfigured LUNs, which include the locks and also management used by Oracle PCA software. And also, there are LUNs and storage used by Oracle VM.
So we have an iSCSI server pool. Now iSCSI server pool, this is the server pool file system, meaning that since so we create clustered server pool, disk heartbeat must be written to disk. So in order to keep the membership of a clustered server pool, we use the server pool file system, and it’s only 12 gigabytes.
And also, we have a default repository. So we have an iSCSI repository. Its name is Rack1_Repository. And that is created out of this LUN. So in the predefined LUN, we create OCFS2 file system, and then we’re going to take it as a default repository.
So we have an OCFS2-based repository, which is called Rack1_ServerPool. And as an option, we can also use NFS-based server pool and also NFS-based repository. And these are also preconfigured as well.
So in the Oracle VM in Oracle PCA, we can have a local repository by taking our local storage. And also, in the ZFS storage appliance, on top of iSCSI repository LUN, we created the OCFS2 repository, which is called Rack1_Repository. And also, you can create external repository. So Rack1_Repository is OCFS2 repository. So this is OCFS2 server pool and OCFS2 repository.
And also, we have iSCSI server pool, and that is in our server pool file system. Rack1_ServerPool is a clustered server pool. So this clustered server pool, it requires the server pool file system. So we use an iSCSI server pool. We create OCFS2 file system on top of it as a server pool file system. So disk heartbeat is written to this location. This is a quick summary of storage layout of Oracle PCA for Oracle VM. So let’s come back to slide.
So OCFS2 Oracle VM– OCFS2 is a clustered file system for Linux– multiple nodes to access files on the same disk at the same time. When you create a repository on a shared disk, an OCFS2 file system is created on the shared disk, just like in the Oracle PCA environment. This does not apply to repositories created on NFS shared.
Now, when you create NFS-based repositories, what happens is NFS file must be available, and then we have to create virtual disk, and then we have to create OCFS2 on the virtual disk. There are multiple file system layers when you create NFS-based repositories. And with Oracle VM, OCFS2 enables controlled access to resources shared among Oracle VM servers. And local repository on an Oracle VM Server uses OCFS2 but cannot be shared. So you can create OCFS2-based repository that could be shared or local.
And here are the list of OCFS2 features– Posix-compliant file system, shared file system support, relink feature– and this is a copy on the write algorithm, so you can create the multiple writable snapshots of regular files– and also OCFS2 support, Sparse file support. So when you create the virtual disk, we can only create a copy of the used blocks by skipping all empty blocks. So these are the OCFS2 features.
And OCFS2 components– OCFS2 software include the core file system, which offers a standard file system interface, and O2CB cluster stack, which supports the shared disk cluster feature. And the sheer the disk cluster feature includes the following– a disk heartbeat to detect live Oracle VM Servers. So when you create an OCSF2 clustered server pool, the server pool file system must be available, and then disk heartbeat is written to the server pool file system.
So network heartbeat for communication between servers– so network heartbeat must be sent to all the members of a server pool. And Distributed Lock Manager ensures only one server at a time can write to a file. For example, you want to perform live migration of a VM. At the same time, another admin attempts to stop the VM that is being migrated.
If two operations can occur at the same time, then we’re going to have a problem. That’s why, in order to protect your VM, in the clustered environment, we use Distributed Lock Manager. So when a admin performs live migration at the same time another admin performs a stopping operation, only the first operation must be done to be able to perform the second operation, not concurrently. So concurrency must be managed in the clustered server pool. So, that way, we can keep consistent image of a VM in the clustered environment.
And OCSF2 cluster services– starting the cluster activates several services and processes on each of the Oracle VM Servers in the cluster. The important services and processes– o2cb service, which is essential to cluster operation, ocfs2 service is responsible for file system operation, ocfs2 net process, which create intra-cluster node communication, and also ocfs2 distributed lock management– so that’s part of a DLM. Do not manually start and stop cluster services. Manually configuring or operating cluster can lead to the cluster failure.
The OCSF2 file system ensures unique read/write access with the Distributed Lock Manager.
Recent Comments